v4.2.0

[C0untFloyd]

Added ability to move input & target faces order
New CLI Arguments override settings
Small UI changes to faceswapping tab
Added mask option and code for restoration of original mouth area
Updated gradio to v4.42.0
Added CLI Arguments --server_share and --cuda_device_id
Added webp image support

Some additional notes:

Arguments
python run.py --server_share starts a public server
python run.py --cuda_device_id 1 would use cuda gpu number 1 instead of the default 0

When using the installer, you need to add the commandline arguments here:

roop-unleashed/installer/windows_run.bat

Line 4 in ad04ab2

set COMMANDLINE_ARGS=

For example it should look like set COMMANDLINE_ARGS=--cuda_device_id 1 if you want to use your 2nd cuda device (0 is the first)

I didn't have the opportunity to test the cuda device selection as I only have 1, I'd be glad about feedback.

Mouth area restoration is hiding in the advanced masking settings. Be aware that this is pasting back all of the targets mouth area and therefore can easily destroy a lot of face similarity. I also set this to be always on for the live-cam.

---

This discussion was created from the release v4.2.0.

[zeeyannosse]

thank you very much!.
i am going to try now.

[12-blk]

Hello
Plessure i need some help.
I'm planning on buying either 3080ti mobile or A5000 mobile for roop . I've found 2 laptops but the A5000 is cheaper than the 3080ti.
Pls advice me which is the best gpu for deepface or programs like roop unleashed.

[zeeyannosse]

great start 👍
Number of CUDA devices: 4 Currently used Id: 0 Device Name: NVIDIA RTX A5000
Using provider [('CUDAExecutionProvider', {'device_id': 0})] - Device:cuda
Running on local URL: http://127.0.0.1:7861

now i need to test in the bat the command python run.py --cuda_device_id 1

[zeeyannosse]

sorry.. i am not a coder. so here is my poor result attempt on the screen grab.
appologies in advance for the naive attempt ;))))

[C0untFloyd]

Hm looks good to me and it should work that way. I need to test this myself when I'm back home.

When using the installer, you need to add the commandline arguments here:

roop-unleashed/installer/windows_run.bat

Line 4 in ad04ab2

set COMMANDLINE_ARGS=

It should look like set COMMANDLINE_ARGS=--cuda_device_id 1 for cuda device 1

[lost-in-emotions]

NICE. will try it out! Thanks a lot!

[VanHellsyng-23]

Hi, I tried to do a clean installation of the program on a Windows 10 operating system. The installation completes successfully, I also installed the ffmpeg libraries and everything else, but when I launch the program, it gives me an error saying that I don't have an NVidia graphics card installed.
Considering that the program has always worked with just the integrated GPU... has something changed? Does it now only work if there's an NVidia card installed on the PC? Is there a way to fix this error?

Thank you very much :)

[C0untFloyd]

Switch the provider in the settings tab to 'cpu' and you should be good.

[VanHellsyng-23]

Sorry, maybe I didn't explain myself well, the issue is that the program doesn't even start. When I try to launch it, I get this error screen.

[C0untFloyd]

Oh I'm sorry, I didn't think about users starting up with cpu-only but cuda enabled. This is bug which I fixed right now:
0627ac9

Thanks for the heads up, please update!

[VanHellsyng-23]

Thanks very much!

[ali-m-github]

Thanks for the update!
I have encountered one issue so far or maybe not an issue but I used to use "open image in new tab" to check images in full resolution and now whenever I use that option it auto downloads the image in .webp format. Is it possible to revert this option back to the preview version? I am using Chrome on Windows 10.

[Elise96nl]

my old .bat file with

python run.py --execution-provider dml

doesnt work anymore, it says no such thing as --execution-provider dml, so i removed it, and just selected dml in the settings menu. and restarted, works ok. so no more selecting it in startup?

[Salimander9]

I usually use the command python run.py --execution-provider dml and it works. Now it is no longer able to run that way and I instead have to launch it the standard way with python run.py. I am able to launch roop however as soon as I interact with anything including changing my provider from cpu to dml I get this error:

pydantic.errors.PydanticSchemaGenerationError: Unable to generate pydantic-core schema for <class 'starlette.requests.Request'>. Set arbitrary_types_allowed=True in the model_config to ignore this error or implement __get_pydantic_core_schema__ on your type to fully support it.

If you got this error by calling handler() within __get_pydantic_core_schema__ then you likely need to call handler.generate_schema(<some type>) since we do not call __get_pydantic_core_schema__ on <some type> otherwise to avoid infinite recursion.

For further information visit https://errors.pydantic.dev/2.9/u/schema-for-unknown-type

I am a noob at coding and unsure of how to fix the issue or where to go.

[Elise96nl]

I think 1 thread, restart, and then switch to DML witouth the -execution-provider dml in the command

[C0untFloyd]

About the pydantic issue:
#886

CLI arguments were removed a year ago or so, from the main page

😄

After updating to the latest version you need to pip install the dml onnxruntime again to make it work.
#186 (comment)

[jp7189]

Testing this out on Win11 with 2x GPUs.
I simply changed this one line in installer\windows_run.bat :
set COMMANDLINE_ARGS=--cuda_device_id 1

It seems to work, and the fails as soon as it hits the first face:
Number of CUDA devices: 2 Currently used Id: 1 Device Name: NVIDIA GeForce RTX 4090
Using provider [('CUDAExecutionProvider', {'device_id': 1})] - Device:cuda

RuntimeError: Error in execution: D:\a_work\1\s\onnxruntime\core\framework\utils.cc:167 onnxruntime::utils::BatchOrCopyMLValue allocator != nullptr was false. Failed to find allocator for device Device:[DeviceType:1 MemoryType:0 DeviceId:0]

Full error message:

Traceback (most recent call last):
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\gradio\queueing.py", line 575, in process_events
response = await route_utils.call_process_api(
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\gradio\route_utils.py", line 322, in call_process_api
output = await app.get_blocks().process_api(
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\gradio\blocks.py", line 1935, in process_api
result = await self.call_function(
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\gradio\blocks.py", line 1532, in call_function
prediction = await utils.async_iteration(iterator)
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\gradio\utils.py", line 671, in async_iteration
return await iterator.anext()
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\gradio\utils.py", line 664, in anext
return await anyio.to_thread.run_sync(
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\anyio\to_thread.py", line 56, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\anyio_backends_asyncio.py", line 2177, in run_sync_in_worker_thread
return await future
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\anyio_backends_asyncio.py", line 859, in run
result = context.run(func, *args)
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\gradio\utils.py", line 647, in run_sync_iterator_async
return next(iterator)
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\gradio\utils.py", line 809, in gen_wrapper
response = next(iterator)
File "C:\StableDiffusion\roop-unleashed\installer\roop-unleashed\ui\tabs\faceswap_tab.py", line 728, in start_swap
batch_process_regular(output_method, list_files_process, mask_engine, clip_text, processing_method == "In-Memory processing", imagemask, restore_original_mouth, num_swap_steps, progress, SELECTED_INPUT_FACE_INDEX)
File "C:\StableDiffusion\roop-unleashed\installer\roop-unleashed\roop\core.py", line 237, in batch_process_regular
batch_process(output_method, files, use_new_method)
File "C:\StableDiffusion\roop-unleashed\installer\roop-unleashed\roop\core.py", line 341, in batch_process
process_mgr.run_batch_inmem(output_method, v.filename, v.finalname, v.startframe, v.endframe, fps,roop.globals.execution_threads)
File "C:\StableDiffusion\roop-unleashed\installer\roop-unleashed\roop\ProcessMgr.py", line 320, in run_batch_inmem
future.result()
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\concurrent\futures_base.py", line 451, in result
return self.__get_result()
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\concurrent\futures_base.py", line 403, in __get_result
raise self._exception
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\concurrent\futures\thread.py", line 58, in run
result = self.fn(*self.args, **self.kwargs)
File "C:\StableDiffusion\roop-unleashed\installer\roop-unleashed\roop\ProcessMgr.py", line 239, in process_videoframes
resimg = self.process_frame(frame)
File "C:\StableDiffusion\roop-unleashed\installer\roop-unleashed\roop\ProcessMgr.py", line 353, in process_frame
num_swapped, temp_frame = self.swap_faces(frame, temp_frame)
File "C:\StableDiffusion\roop-unleashed\installer\roop-unleashed\roop\ProcessMgr.py", line 447, in swap_faces
temp_frame = self.process_face(self.options.selected_index, face, temp_frame)
File "C:\StableDiffusion\roop-unleashed\installer\roop-unleashed\roop\ProcessMgr.py", line 609, in process_face
sliced_frame = p.Run(inputface, target_face, sliced_frame)
File "C:\StableDiffusion\roop-unleashed\installer\roop-unleashed\roop\processors\FaceSwapInsightFace.py", line 48, in Run
self.model_swap_insightface.run_with_iobinding(io_binding)
File "C:\StableDiffusion\roop-unleashed\installer\installer_files\env\lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 331, in run_with_iobinding
self._sess.run_with_iobinding(iobinding._iobinding, run_options)
RuntimeError: Error in execution: D:\a_work\1\s\onnxruntime\core\framework\utils.cc:167 onnxruntime::utils::BatchOrCopyMLValue allocator != nullptr was false. Failed to find allocator for device Device:[DeviceType:1 MemoryType:0 DeviceId:0]